PDTF to Ontology: Programme and Work Breakdown (Anchor)

Context

This is the anchor record for converting the Property Data Trust Framework v3 JSON Schema into a linked-data ontology. It records the programme-level decisions from Council Session 001, sequences the work, and links every work-package ODR. It is planning only — each linked ODR is a stub to be fleshed out in its own follow-up session.

The conversion problem is not a mechanical Schema-to-RDF rewrite. The web-app schema section already names the load-bearing defect (the implicit Property entity, page 37): UPRN appears in four leaf paths, address in many, an INSPIRE ID and a title-linked address besides, with zero schema-level joins between them — a missing class with no identity criterion. JSON Schema gives slot-names, not global identifiers, so the genuine modelling question is which things get URIs. The mechanical half (named slot → DatatypeProperty with xsd: range) is generated; Council cycles are reserved for the ambiguous moves (aggregate boundaries, cross-overlay synonymy, oneOf-as-subclass-vs-state).

Inputs converted: pdtf-transaction.json (37,224 lines, JSON Schema Draft-07) — base residential-property-transaction model for England & Wales plus the verifiedClaims OIDC4IDA/eIDAS envelope and 10+ deep-merge form overlays (BASPI, TA6/7/10, NTS, LPE1, CON29R/DW, LLC1, FME1); the web-app schema section (11 pages, 3,561 leaves walked, 15 overlays cross-referenced); the PDTF business glossary (54 trust-framework/open-banking terms merged with schema annotations and W3C VC / DID Core / ToIP inheritances) supplying authoritative definitions for rdfs:label / skos:prefLabel / skos:definition; the PDTF data dictionary (1,557 unique leaves across 8,458 path entries; 935 of 1,556 base leaves carry semantic annotation; per-form counts — baspi5 318, rds 196, piq 184, ta6 178, nts2 160, lpe1 136, con29R 125, ntsl2 124, ta7 98, ta10 90, fme1 78, oc1 68, con29DW 34, sr24 7, llc1 3) and the role-enum members that become SKOS schemes. The term-sourcing convention (how each label/definition carries dct:source back to its glossary row or schema leaf path) is owned by ODR-0004, not here.

Convening constraints for Session 001: data model only (TBox; no instance-data deliverable, later amended to admit diagnostic exemplars); vocabulary floor of Core + DASH + PROV-O + the data-governance family; BBO and other non-relevant Conditional vocabularies excluded; output a set of work-partitioning ODRs plus this anchor.

Decision

Partition the PDTF→ontology programme by ontological concern (FIBO-style modules reconciled with Guizzardi’s UFO Kind/Role/Relator layering), not by aggregate page, because mirroring the JSON tree encodes form ergonomics rather than ontological cohesion; sequence the work spike-then-scale with the Property identity criterion as the gating crux, prove the pipeline end-to-end on one BASPI5 vertical slice, then scale the remaining overlays and modules. This supersedes the by-aggregate-page breakdown of the earlier placeholder stubs.

Rules

Target versions. RDF 1.2 and SHACL 1.2, per the Core-tier pin in ODR-0002.

Programme-level decisions adopted from Session 001.

Supersession scope. This anchor supersedes the by-aggregate-page placeholder stubs that preceded it; the concern partition replaces them. Each linked ODR owns its own analysis.

Work breakdown.

The diagram below shows the four programme phases and the ODRs they contain, with hard gates between them.

Phase 0 — Spike (gates the programme):

• ODR-0004 — Foundation. URI/namespace strategy (single opda: hash namespace), ontology-header pattern, OWL-graph ⊥ SHACL-graph separation, generator-first policy, diagnostic-exemplar policy, and the term-sourcing / dct:source convention drawing on the business glossary and data dictionary.
• ODR-0005 — Property & Land identity crux. The gate. Class split, DOLCE Endurant commitment, identity criteria over demolition/subdivision/merger/first-registration, UPRN key-vs-contingent-identifier resolution — validated against diagnostic exemplars. No module ODR is drafted in anger until this clears.

Phase 2.6 — Address gate (between IC gate and Agents, per Scope-Check 1 Q7a):

• ODR-0015 — Address & Geography. Declares opda:Address as a first-class endurant distinct from opda:Property; settles UFO category (Kind/Quale/Mode); INSPIRE Identifier and UPRN-as-geographic-identifier; GeoSPARQL deferral home. Consumed by 006, 008, 009, 012. Gate before Sessions 006 and 008 (added per Scope-Check 1 Q7a, 8-1 mandatory spawn).

Phase 1 — Modules (after the crux + Address clears):

• ODR-0006 — Agents & Roles. Person/Organisation Kinds; Seller/Buyer RoleMixins; Proprietor Role + Proprietorship Relator; capacity-vs-evidenced-authority; FOAF ruled out (Kind-layer vocabulary now W3C Org vs bespoke opda:). Reuses opda:Address from ODR-0015.
• ODR-0007 — Transactions & Lifecycle. Transaction relator, milestones, status; OWL-Time intervals.
• ODR-0008 — Property descriptive attributes. Built form, condition, valuation, EPC/energy, utilities, local-context searches, encumbrances/completion. Reuses opda:Address from ODR-0015. (Sub-module split deferred per Scope-Check 1 Q2 vote 2-7; two candidate splits with named triggers recorded in plan §11.)

Cross-cutting (drafted alongside, after ≥1 module exists):

• ODR-0009 — Claims, Evidence & Provenance. PROV-O backbone + assurance layer. DPV co-annotation authoring moved to ODR-0012 per Scope-Check 1 Q5 refinement; ODR-0009 carries a one-paragraph pointer.
• ODR-0010 — Overlay Profile Mechanism. SHACL profiles, opda:ValidationContext, dct:source traceability, DASH rendering. Cross-cites ODR-0013 on three interface rules (sh:in semantics; sh:Violation floor; no-identity-override gate) per Scope-Check 1 Q6 (Cagle’s three-rule contract).
• ODR-0011 — Enumeration Vocabularies. JSON enums → SKOS concept schemes. Each scheme declares its UFO meta-category (Quale-in-Region / Role label / Phase label / method-plan code) per Scope-Check 1 Q3 (Guizzardi sub-finding).
• ODR-0012 — Data-Governance Layer. DPV Phase-1 annotation (+ Pandit’s recorded dissent), ODRL deferred. Owns DPV co-annotation authoring (ODR-0009 cites).
• ODR-0013 — SHACL Validation & Severity. Constraint mapping, severity tiering, DASH UI, annotation-graph separation. Cross-cites ODR-0010 on the three interface rules above.

Phase 7 — Deferred (named-but-not-running):

• ODR-0016 — W3C Verifiable Credentials / DID Compatibility Layer. Named per Scope-Check 1 Q7c (vote 8-1). Deferred — activates on any of: session-009 Q8 surfaces real VC-side decisions; session-012 Phase-2 consent receipts land; a real wallet / DID consumer enters scope. cred: and did: prefixes admitted to ODR-0002’s Defer tier immediately.

Retired:

• ODR-0014 — Retired by Scope-Check 1 Q4 (vote 7-1-1; Hendler dissent on permanence recorded). Amendment rows folded into ODR-0002’s ## Change log. ODR-0014 retained as historical anchor.

Phase ordering authority (added per Session 003 Item 1).

The canonical phase sequence is the Council follow-up sessions plan §5. The session table within the plan is not reproduced here (single source of truth). The phase order is binding until amended: a session cannot unilaterally re-sequence. Substantive re-cuts land as a joint amendment to both the plan and this anchor in one commit. Pre-Phase Session A9 (Gandon-Guizzardi methodology gap; Reduced Council; recommended before Session 005) and the two pilot sessions (S005 with consensus-mode: hive-mind/byzantine per Scope-Check 2 B2; S011 Q8 with consensus-mode: hive-mind/typed-output per Scope-Check 2 B3) are named in the plan.

Module count. Post-Scope-Check 1: 3 modules + 5 cross-cutting + 1 substrate + 1 new gate + 1 deferred = 11 active ODRs (plus retired ODR-0014, retained as historical anchor):

• Modules (Phase 3a/3b): ODR-0006, ODR-0007, ODR-0008.
• Cross-cutting: ODR-0009, ODR-0010, ODR-0011 (promoted to substrate Phase 2.5), ODR-0012, ODR-0013.
• Substrate (Phase 0): ODR-0002 (absorbs former ODR-0014 amendments per Scope-Check 1 Q4).
• New gate (Phase 2.6): ODR-0015 (spawned by Scope-Check 1 Q7a). Address class location is owned by Session 015, not by Session 006 (Q5).
• Deferred (Phase 7): ODR-0016 (named per Scope-Check 1 Q7c; activates on trigger).

Dependency graph (updated per Scope-Check 1, 2026-05-26).

                    ODR-0004 Foundation
                          │
                          ▼
              ODR-0005 Property identity CRUX ◀── diagnostic exemplars (Q1)
                          │  (GATE — must clear before Phase 1)
                          ▼
              ODR-0015 Address & Geography ◀── new gate (Scope-Check 1 Q7a)
                          │  (GATE — must clear before Sessions 006 and 008)
        ┌─────────────────┼─────────────────┐
        ▼                 ▼                 ▼
   ODR-0006          ODR-0007          ODR-0008
   Agents&Roles      Transactions      Property attrs
        └─────────────────┼─────────────────┘
                          ▼
   Cross-cutting (need ≥1 module):
   ODR-0009 Claims/Provenance ──▶ ODR-0012 Governance
            (DPV co-annotation authoring owned by 0012; 0009 cites)
   ODR-0010 Overlay profiles ◀═══cross-cite═══▶ ODR-0013 SHACL/severity
            (three interface rules per Scope-Check 1 Q6 / Cagle)
   ODR-0011 Enumerations (each scheme declares UFO meta-category)

   ODR-0016 W3C VC/DID — deferred-named; activates on triggers (see above).
   ODR-0014 — RETIRED; folded into ODR-0002's ## Change log.

The Mermaid diagram below renders the same dependency graph with cross-cite edges shown explicitly.

Shared-question routing (added per Session 003 Item 7).

Several questions surface in more than one session. Each shared question is owned by one session; downstream sessions inherit. Routing is maintained in plan §4.1 as the single source of truth (full table not duplicated here). If a downstream session genuinely needs to deviate from the owning session’s verdict, it records the deviation as a ## Supersession scope: amendment on the owning ODR’s ## Rules. Routing failures (two sessions both producing a verdict on the same shared question) are a defect — the later session’s verdict is invalid pending an explicit amendment cycle. Notable cross-cite: ODR-0010 ↔ ODR-0013 carry the three-rule SHACL interface contract (sh:in semantics; sh:Violation floor; no-identity-override gate) per Scope-Check 1 Q6 (Cagle).

Minimum viable subset. Foundation (0004) → Property identity crux (0005, exemplar-gated) → Address & Geography (0015, gate before 006/008) → Agents & Roles (0006) + Claims/Provenance (0009) → one fully-worked BASPI5 SHACL profile (0010). The intra-MVP ordering places the PROV-O claims/assurance backbone before the profile scale-out (provenance is foundational to a trust framework and higher integrity-risk than a form overlay). See plan §5.1 for the MVP fast-path option.

Default sequence vs MVP fast-path (added per Session 003 Item 2).

Two named, first-class alternatives:

• Default sequence (plan §5) — ratify all 13 active sessions plus deferred S016 before implementation begins. Cost: ~10–14 working days; ~87 agent runs in Phase 1.
• MVP fast-path (plan §5.1) — ratify only the minimum subset (S003 / S002 / S004 / S005 / S011-light / S015 / S006 abridged / S009 PROV-O only / S010-BASPI5) to reach the BASPI5 round-trip handoff; resume the remaining sessions in parallel with or after implementation. Cost: ~6–8 working days; ~50 agent runs to handoff. Pays in two re-open cycles (S006 Q6 + S011 full) and deferred completeness on S007/S008/S012/S013.

The OPDA Working Group chooses between these at the start of Phase 1; the choice is recorded in this anchor by an Author-only Session 003b when made. Switching after Phase 1 starts is expensive and discouraged.

Gate conditions (enforcement).

• The identity crux clears first. ODR-0005 must (i) commit each property/title entity to a DOLCE category (Endurant), (ii) state an identity criterion over the hard cases (demolition / subdivision / merger / first-registration), and (iii) settle UPRN’s status (checkable SHACL/DASH key vs contingent administrative identifier) — all validated against the diagnostic exemplars (registered freehold house; unregistered house pre-first-registration; flat whose UPRN was split). No module ODR (0006–0008) is drafted in anger until this clears.
• MVP round-trip gate. The minimum-viable subset must round-trip one BASPI5 profile: pdtf-transaction.json → loaded SHACL profile → rendered BASPI form (via DASH) → validated provenance (PROV-O claims slice) with full dct:source traceability. If that round-trips, the remaining overlays and modules scale after. Termination signal 1 is now demonstrated by the ci-baspi5-roundtrip gate (ADR-0014): loading base shapes + the BASPI5 overlay validates a conformant transaction (no violations), reports a violation on a non-conformant one (a Seller acting as Attorney with no evidenced authority, the sh:xone sellersCapacity branch, traceable via dct:source to form-question B1.3.2), and confirms every field carries a resolvable dct:source + DASH render hint. The actual DASH-UI render is the consumer-app boundary; opda-gen guarantees the round-trip data contract (every field → resolvable form-question + render hint), not the UI.

Programme retirement criterion (added per Session 003 Item 5).

The programme retires when both hold: (i) the MVP round-trip closes (termination signal 1 per plan §5 — pdtf-transaction.json → loaded SHACL profile → rendered BASPI form via DASH → validated transaction with dct:source traceability); and (ii) every linked ODR (the active 11 plus retired ODR-0014’s historical anchor; ODR-0016 only if its trigger has fired) is accepted. Termination signals 3–6 (no duplicate constraint authoring; ≤3-ODR consumer-query traversal; ODR-0003 diff stability; PII never accretes silently) are cumulative quality gates evaluated at session close, not retirement conditions — their violation routes back to the ODR-0001 amendment queue.

Once retired, ODR-0003 becomes a historical anchor; subsequent linked-data modelling work in OPDA produces fresh ODRs without revisiting this programme’s sequencing.

Status discipline (bidirectional-update protocol) (tightened per Session 003 Item 6).

This anchor is the single place to see programme state. The queen of the session that ratifies an ODR (proposed → accepted) is responsible for three bidirectional updates in the same commit as the session transcript and the ODR amendment:

1. This anchor’s ## Rules work-breakdown index — flip the relevant Phase entry’s status pointer; note cascading effects on downstream sessions if any.
2. The OPDA adoption record §Track Record table — add the session row (date, session ID, format, Queen, DA, subject, one-line verdict).
3. Session transcript and ODR cross-references — verify the ODR’s ## References links the transcript; verify the transcript header links the ODR; run odr-review to lint frontmatter, section structure, and referential integrity; run odr-index to refresh AgentDB graph edges.

Pilot sessions (S005, S011 Q8) carry an additional artefact per plan §8: a one-page retire-or-extend evaluation in the transcript; the outcome (RETIRE / EXTEND CAUTIOUSLY / EXPAND) is recorded in the track-record row’s verdict column.

Individual ODRs own their own analysis; this file owns the sequencing and the cross-links.

Alternatives

The diagram below summarises the three rejected partition strategies and why the chosen option was preferred.

• Partition by aggregate page — mirror the JSON tree and the web-app’s 11 schema pages, one module per aggregate. Rejected: encodes form ergonomics, not ontological cohesion; duplicates reused entities (Address, Name, Person, Organisation) across modules; treats Evidence and VerifiedClaims as siloed pages when they are cross-cutting relations.
• Partition by UFO meta-category alone — Substance Kinds / Roles & Phases / Relators & Claims. Rejected as a sole cut: too coarse to map onto FIBO-style concern modules; subsumed into the chosen option as its layering axis.
• Drive the whole 15-ODR programme up front (no spike) — Rejected: the identity-criterion question is too contested to draft module ODRs against; spike-then-scale front-loads the hard constructs into one proven vertical slice.

Consequences

• Declare reused entities (Address, Name, Person, Organisation) once, not per aggregate page, and isolate open-world class semantics from closed-world shape validation.
• Promote Evidence/Claims/Enumerations/Governance/Validation to cross-cutting status; model them as the relations they are rather than forcing them into module silos.
• Keep the published namespace flat; module structure is editorial, so re-grouping concepts later does not break dereferenceable URIs.
• Front-load the genuinely hard constructs (identity criteria, sh:xone, capacity, DASH editors, dct:source) into a single proven vertical slice before the largely-mechanical overlay scale-out.
• Treat the identity crux (ODR-0005) as a hard single-point gate early in the programme: module ODRs (0006–0008) are not drafted in anger until it clears.
• Re-cut any work already premised on the superseded by-aggregate-page placeholder stubs to the concern partition.
• Update this anchor as ODRs progress; do not duplicate per-ODR analysis here.

References

• Council methodology: ODR-0001.
• Vocabulary catalogue: ODR-0002 (amendments folded inline; ODR-0014 retired per Scope-Check 1 Q4).
• Phase 0 spikes: ODR-0004, ODR-0005.
• Phase 2.6 gate: ODR-0015 — Address & Geography (added per Scope-Check 1 Q7a).
• Phase 1 modules: ODR-0006, ODR-0007, ODR-0008.
• Cross-cutting: ODR-0009, ODR-0010, ODR-0011, ODR-0012, ODR-0013.
• Phase 7 deferred: ODR-0016 — W3C VC / DID Compatibility Layer (named per Scope-Check 1 Q7c; activation triggered).
• Retired: ODR-0014 — folded into ODR-0002 per Scope-Check 1 Q4.
• Deliberation provenance — Session 001: session-001-pdtf-schema-to-ontology. Full per-question positions, vote tallies, recorded dissents (including Guarino’s Devil’s-Advocate scorecard and DA withdrawals) live there; this anchor records sequencing and cross-links only.
• Ratification provenance — Session 003 (Author-only; 2026-05-27; Queen Kendall): session-003-pdtf-ontology-programme. Records phase ordering (plan §5), default-vs-fast-path option (plan §5.1), identity-crux gate check, module count, programme retirement criterion, status-discipline bidirectional-update protocol, and shared-question routing (plan §4.1) into the ## Rules above. No fresh deliberation claimed; every item plan- or precedent-sourced.
• Programme-level scope review: Scope-Check 1 — Programme cut (2026-05-26, Queen Kendall, DA Davis). Verdict 8-1 APPROVE the cut with nine named amendments; spawned ODR-0015, named ODR-0016, retired ODR-0014, moved DPV co-annotation authoring to 0012, added Cagle’s three-rule interface contract between 0010 and 0013, recorded Guizzardi’s UFO-per-scheme sub-finding for 0011, surfaced Gandon-vs-Guizzardi methodology gap (routed to ODR-0001 amendment queue).
• Methodology-tooling review: Scope-Check 2 — Hive-mind vs Agent fan-out (2026-05-26, Queen Kendall, DA Davis). Verdict 5-1 SELECTIVE; Agent fan-out stays default. Two pilot sessions named for ruflo hive-mind consensus: S005 (Identity crux) with consensus-mode: hive-mind/byzantine (tests cross-conditional voting hypothesis), and S011 Q8 (UFO meta-category per scheme) with consensus-mode: hive-mind/typed-output (tests typed downstream consumption hypothesis). Scope-Check 2 B1 (consensus-mode framework) landed in ODR-0001 via direct amendment 2026-05-27 (Author-only self-amendment); pilots unblocked. A9 (Scope-Check 1 — Gandon-Guizzardi methodology gap) still pending but not strictly blocking — recommended before S005. Pilots are an evaluation budget, not an adoption commitment — retire-or-extend decision at session close.
• Project adoption context: OPDA Council adoption record — declares OPDA’s project-specific instantiation of the methodology (panel weighting, pre-elected extended panel, governance handoff, track record, when-to-use additions, council directory path). Per ODR-0001’s portable methodology design, project specifics live in the adoption record, not in the methodology body.
• Follow-up programme execution: Council follow-up sessions — operationalises the work-breakdown above by attaching one Council session to each linked stub, with dependency-ordered phasing and explicit gates at ODR-0004 (Foundation), ODR-0005 (Identity crux), and ODR-0015 (Address).
• Source inputs: pdtf-transaction.json; web-app schema section (src/pages/schema/*.astro, source/_content/schema/*.md); PDTF business glossary (source/00-deliverables/semantic-models/business-glossary.md); PDTF data dictionary (source/00-deliverables/semantic-models/data-dictionary.md).